The transport properties of porous materials have been a subject of scientific interest for many years, and have become the focus of much attention in the last 10-20 years, primarily due to work carried out in the oil-well logging and exploration community on porous sedimentary rocks. There the interest has been in predicting transport coefficients based on microstructural parameters of the pore space. These transport coefficients include the electrical conductivity of the pore space [1], the diffusivity of the pore space, which is related to the conductivity via an Einstein relation [2,3], and the fluid permeability [4].
Cement-based materials are also porous materials whose transport coefficients are of interest, but for different reasons than for rocks. The focus of interest in rocks has been fluid permeability, which is not unreasonable, since oil is a fluid that must be pumped through and out of the porous rocks in which it is found. In cement-based materials, the transport of dissolved chemical species through the pore space is of more significance, for the following reasons. First, most of the physicochemical processes that degrade cement-based materials and ultimately determine service life depend on a supply of ionic species from external sources [5]. The rate at which these species can move through the pore structure largely determines the rate at which degradation proceeds. Examples include chloride ions attacking reinforcing steel in concrete, and sulfate ions reacting with various phases in concrete to produce crack-causing internal expansive pressures [6]. Second, there has been much recent interest in using cement-based materials to contain low and intermediate level radioactive and also toxic waste [7]. The transport coefficients of these materials are the key factors that will determine their effectiveness as barriers.
Although both fluid permeability and ionic diffusivity are
important transport coefficients for cement-based materials, this
paper focuses on the diffusivity. A digital-image-based growth
model of the developing microstructure of cement paste during
hydration is coupled with two algorithms for computing the
conductance of random conductor networks in order to carry out
the computations to be described below. Preliminary accounts of
some of this work have previously appeared [8,9].